-
If we have the following model
```
model
schema 1.1
type user
type document
relations
define a: [user]
define b: [user]
define b_rewrite: b
define c: a o…
-
Hi, thanks again for this excellent repo. The model trains fine for me on a multi-gpu system using deepspeed stage 2 with or without offload, and I see the expected training/validation time reduction…
-
This also makes `data_summary` and `Base.show` very slow since showing a `FieldTimeSeries` prints its min, mean, and max. So it's harder to work with `FieldTimeSeries` interactively. Seems fine when n…
-
1. When a file with the same name as the output file already exists in the output folder.
2. When using the waifu2x model, select "none" as the noise reduction parameter.
Yipky updated
3 weeks ago
-
### 🚀 The feature, motivation and pitch
I propose implementing int8 quantization support for vLLM, focusing initially on the KV cache. This feature will allow users to run larger models or increase b…
-
Hi I am getting decent enough results on the vit-s architecture model for metric depth which shows it has around 22-25 million parameters. I want to make the model even faster/lighter, I could think o…
-
## In pytorch
...
norm = torch.norm(x, p=2)
return x / norm
I converted this model into onnx and tried to convert to ncnn, got error "Unsupported reduction axes !"
How to pass model conversion …
-
**Describe the bug**
While using the ndarray backend, if the input to LayerNorm is a zero vector, it results in NaN during backward propagation.
**To Reproduce**
The reproduction code is as follo…
-
Hey, guys. I am trying to train the flow model from scratch recently.
But I am a bit confused about the training pipeline of the flow model. **As suggested by [#281 @aluminumbox](https://github.co…
-
Hi,
Are there plans to add the ability to model pile groups, where the pile cap is assumed to be rigid?
I gather the main additions to the code would be:
- Adding a rigid link between the tops of th…